My own product, 2026 – present
Aaina Dekho: AI virtual try-on for South Asian fashion
Shoppers upload one photo and see themselves in a kameez, dupatta or saree. Fashion brands add the same try-on to their product pages with one script tag.
Context
Online fashion shoppers in Pakistan cannot see how an outfit looks on them, and generic try-on tools handle dupattas, hijabs and embroidery badly.
My role
Founder and sole engineer.
What I did
- Built the try-on app in Angular with standalone components and signals: a model photo from one selfie, try-on, outfit layering, pose changes and a before/after view.
- Wrote prompts specifically for South Asian garments and embroidery, on Google Gemini image models with an automatic fallback model.
- Built serverless API functions on Vercel with layered abuse protection: per-device, per-IP, per-key and global daily limits counted in Redis.
- Made an embeddable "Try it on" widget that brands add with one script tag, with keys checked against the brand’s domain.
- Covered the product with about 86 Playwright end-to-end tests and 48 API unit tests, and added guide pages in English and Urdu for search.
Outcome
- Live at aainadekho.com. The brand widget is in a free beta.
Need something similar?
- Remote Angular developmentA senior Angular engineer joining your team remotely to ship features end to end.